One form and two functions: MBD of SETDB2 is a protein-interacting domain

Abstract

In this issue of Structure, Mahana et al.¹ present their structural characterization of an annotated methyl-CpG-binding domain (MBD) from the histone H3 lysine 9 methyltransferase SETDB2. This study reveals that, rather than binding DNA as previously hypothesized, this domain instead interacts with a cystine-rich domain from C11orf46, highlighting its involvement in protein-protein interactions.

“Form follows function”—that has been misunderstood. Form and function should be one, joined in a spiritual union.

Frank Lloyd Wright

Histone H3 lysine 9 (H3K9) methylation, a repressive epigenetic modification, is catalyzed by specific methyltransferases (MTases). Among the earliest characterized histone lysine MTases were SUV39H1 in humans and Suv39h1 in mice.² Subsequent work revealed three pairs of H3K9 MTases in mammals (Figure 1): (1) SUV39H1 and SUV39H2, which predominantly generate tri- or dimethylated H3K9 (H3K9me3/me2) within heterochromatin regions; (2) EHMT1 (euchromatic histone lysine MTase 1 or G9a-like protein, GLP) and EHMT2 (commonly referred to as G9a), responsible for creating di- or mono-methylated H3K9 (H3K9me2/me1) in euchromatin; and (3) SETDB1 and SETDB2 (SET domain bifurcated histone lysine MTase 1 and 2), which also produce H3K9me3/me2 silencing marks. These marks are crucial for repressing transposable elements, repetitive sequences like endogenous retroviruses, and pericentric satellite repeats.

Figure 1. Schematic diagrams of the three pairs of human H3K9-specific MTases discussed in this preview.

The protein lengths and their corresponding UniProt accession numbers are shown.

(A) SUV39H1/H2.

(B) EHMT1/GLP and EHMT2/G9a. “Unknown” refers to a partially folded domain predicted by AlphaFold.

(C) SETDB1/2 and C11orf46.

The catalytic SET domain, named after Su(var)3–9, Enhancer-of-zeste, and Trithorax, is illustrated in green. In SETDB1/2, the corresponding SET domain is split or “bifurcated” by a significant insertion sequence, the function of which is yet to be determined. All these H3K9-specific MTases feature at least one additional protein-protein interacting module. The chromodomain in SUV39H1/2 and the ankyrin repeat domain in G9a/GLP serve as histone methyl lysine binding domains. The “MBD” (colored in orange) in SETDB1/2 engages with the zinc-binding cysteine-rich domain (CRD) of C11orf46. Additionally, the tandem Tudor domain in SETDB1 and the N-terminal domain of C11orf46 are implicated in binding to modified histone tails (H3K14 acetylation and H3R2 citrullination).

(D) Sequence alignment of CRD of C11orf46 with two potential orthologs in A. vaga (accession # UJR22702 and UJR26342). The zinc coordinating cysteine residues are highlighted white on a red background. Invariant residues across the three proteins are shown white on a black background, and conserved residues (K and R; V, L and I; W and Y; and N and Q) are indicated with a gray background. Notably, the similarity among these sequences extends beyond the CRD.

In addition to their catalytic SET domains, H3K9-specific MTases possess at least one other distinct protein module within their sequences. For instance, SUV39H1/H2 feature a chromodomain, G9a/GLP contain an ankyrin domain, and SETDB1/2 are equipped with an MBD (methyl-CpG-binding domain) (Figures 1A and 1B). The chromodomain in SUV39H1 recognizes histone peptides that contain H3K9me3/2,³ whereas the ankyrin repeats in G9a and GLP bind to H3K9me2/1.⁴ This signifies that SUV39H1/H2 and G9a/GLP represent types of histone MTases that contain both the machinery to generate and to recognize the same epigenetic mark within a single polypeptide chain, thereby facilitating the spread of the silencing mark across a specific chromatin region. However, such a dual-function capability—both creating and binding a specific histone mark (H3K9me3/2)—does not appear to be present in SETDB1/2.

In addition to its MBD, SETDB1 also features an N-terminal tandem Tudor domain (Figure 1C). This Tudor domain does not exhibit binding affinity for H3K9 methylation (H3K9me3/2/1). Instead, it binds to peptides that are both methylated at H3K9 and acetylated at H3K14, albeit with a low affinity (in the high micromolar range).⁵ The significance of the coexistence of these modifications—acetylation at H3K14, which is typically associated with transcriptionally active, open chromatin, and methylation at H3K9, a mark of repression—remains to be fully understood. However, the presence of a putative MBD in both SETDB1 and SETDB2 suggests a compelling model that connects H3K9 methylation with DNA methylation, the latter being a well-established epigenetic repression mechanism. This raises intriguing possibilities about the coordination of histone and DNA modifications in regulating gene expression and chromatin states. Specifically, DNA methylation is often associated with the absence of H3K4 methylation and the presence of H3K9 methylation. The interactions between (1) DNMT1 and G9a⁶ and (2) DNMT3A and G9a/GLP, mediated by MPP8,⁷ provide some molecular insight into the co-occurrence of DNA methylation and H3K9 methylation in chromatin. The integration of a DNA methylation “reader” (MBD) with an H3K9me3 “writer” in SETDB1 and SETDB2 raises an intriguing question: does this intrinsic or associated coupling suggest a mechanism for the propagation and/or maintenance of these two epigenetic marks across the genome? In the current issue of Structure, Mahana et al. provide an alternative mechanism.¹

Rather than binding to DNA or methylated DNA, the MBDs of both SETDB1 and SETDB2 interact with C11orf46, a nuclear protein also referred to as Arl14ep (ADP-ribosylation factor-like 14 effector protein). This 260-residue protein is encoded in a specific region of chromosome 11p13, which is linked to WAGR syndrome (Wilms’ tumor, Aniridia, Genitourinary anomalies, and a range of developmental delays). The MBD of SETDB2 is structurally similar to that of MeCP2—a well-established DNA-binding protein that prefers 5-methylcytosine within a CpG dinucleotide⁸—with an N-terminal extension that coordinates a zinc ion. Although both MeCP2 and SETDB2 MBD domains possess a conserved basic surface, the former binds to methylated DNA, while the latter associates with C11orf46. It is the C-terminal cysteine-rich domain of C11orf46, capable of binding two zinc ions, that attaches to the MBD of SETDB2 (Figure 1C). Consequently, the heterodimer formed by SETDB2 and C11orf46 exhibits three sets of zinc coordination sites.¹

While the study by Mahana et al.¹ clarified that the MBD of SETDB1 and SETDB2 does not bind to DNA, the relevance of their interaction with C11orf46 within the H3K9 methylation-dependent repression complex remains unclear. A previous study⁹ indicated that both full-length C11orf46 and C11orf46 with its C-terminal cysteine-rich domain deleted exhibit stronger binding to histone H3 peptides when they were citrullinated at arginine 2 (H3R2). Nonetheless, the broader biological implications of H3R2 citrullination, especially in relation to the H3K9 methylation mediated by SETDB1/2, continue to be a subject of investigation. Interestingly, AlphaFold modeling of C11orf46 predicts that its N-terminal domain comprises two CCCH motifs, bearing a weak resemblance to the THAP (Thanatos-associated protein)-family zinc coordinating DNA-binding proteins that are involved in various eukaryotic cellular functions.

Although SETDB2 functions remain understudied, SETDB1 has been implicated in a variety of developmental and pathological processes, including embryo-genesis, germ cell development, stem cell behavior, tumorigenesis, and inflammatory bowel disease (see references in¹). The work by Mahana et al.¹ could facilitate further mechanistic studies of SETDB1/2 in these processes. The discovery of an interaction between the MBD of SETDB1/2 and C11orf46 in humans also has broader implications, extending to species in which orthologs of SETDB1/2 and C11orf46 exist. This includes organisms like Drosophila melanogaster and Caenorhabditis elegans that have no or low levels of DNA cytosine methylation in their genomes. Furthermore, the relevance of this finding is also evident in the bdelloid rotifer Adineta vaga, an organism that features DNA N4-methylcytosine rather than the more commonly known 5-methylcytosine.¹⁰ We identified two potential orthologs in A. vaga that possess cysteine-rich domains similar to the domain in C11orf46 (Figure 1D). This indicates a potentially fundamental role for the interaction between SETDB2 MBD and C11orf46 in a variety of biological contexts, even in species with differing DNA methylation modifications.